Arka Sadhu

Hi @zhaoyue-zephyrus Thanks for the amazing repository. I tried with the default settings, but I use 4 gpus. I trained for ~ 2.5 days (12 epochs) with learning rate halved....

@zhaoyue-zephyrus I tried with more epochs (another 10 epochs), the result is still the same map ~ 0.03. I also noticed the following in the log just before report of...

Yes, I used the testing code and using your pretrained model from the model zoo I get 21 map. `PascalBoxes_Precision/[email protected]= 0.21313359468022483` I am using 4 gpus, each being a 1080Ti....

@tdurieux Thank you for the amazing work. Wondering if is there any interest in supporting cloning from an anonymous repository? One use case is where we want to upload our...

@rom1504 Ah, interesting. I wasn't aware of video2numpy project. If I understand correctly, it would store it individual frames but that would result in very large files. Webdataset has some...

Currently, codebases like pytorchvideo are using pyav, ffmpeg or decord to extract raw frames from mp4 files. Training on kinetics requires high fps (30) for models such as SlowFast.

Here is an example ``` """ main.py """ import fire from utils import get_data def main(exp_name, batch_size, num_workers): print(f'Starting Experiment {exp_name}') data = get_data(batch_size, num_workers) if __name__ == '__main__': fire.Fire(main)...

Yeah, I meant this to be an addition. Here is a small example I am cooking up from your example. https://github.com/williamFalcon/test-tube/blob/master/examples/pytorch_hpc_example.py The given code: ``` # set up our argparser...

I only used `%time` for the line `vl = de.VideoLoader(videos, ctx=ctx, shape=shape, interval=interval, skip=skip, shuffle=0)`. If I understand correctly (let me know if I am incorrect), that shouldn't require transfer...

This is interesting. Do you have per-epoch timings? That would be easier to compare I think.